[
Lists Home |
Date Index |
Thread Index
]
- From: "Ingo Macherius" <macherius@darmstadt.gmd.de>
- To: "Michel Rodriguez" <mrodrigu@ieee.org>, "gopi" <gopi@aztecsoft.com>
- Date: Tue, 28 Mar 2000 14:03:18 +0200
> From: owner-xml-dev@xml.org [mailto:owner-xml-dev@xml.org]On Behalf Of
> Michel Rodriguez
> I have just one question regarding APIs for xml storage: why can the DOM
> not be used at one?
The DOM can be used, but it should not. While being the prevalent model
today, recent W3C standards such as XML Schema, XML Query, and XInclude
start to use the XML Infoset rather then DOM-like datamodels.
The reason for this is the perception that data model and the operations
(DOM people read: API methods) to manipulate it should be independent. It is
sufficient to describe what changes on an Infoset a processing step implies,
but leave it open (and optimize able) how it is exactly done. Roughly
speaking, the Infoset is a DOM2 without fixed API.
In DOM thinking, the only way to describe changes in the data model is to
describe the functional composition of DOM API methods that perform the
change. This simply is too low a level of abstraction.
> Does it need extensions? Lower level methods?
No, it needs no methods at all to do the job. Keep data model and API
separated. When it comes to implementation, however, I also do not see a
real alternative to the DOM today. But this must not preclude the freedom of
thinking in terms of the Infoset. Thus: think Infoset, and then implement
DOM what you came up with in as DOM method calls. Or in an optimized way ...
this is where database technology can intrude the XML word :)
Regards,
++im
***************************************************************************
This is xml-dev, the mailing list for XML developers.
To unsubscribe, mailto:majordomo@xml.org&BODY=unsubscribe%20xml-dev
List archives are available at http://xml.org/archives/xml-dev/
***************************************************************************
|
